GPT4 and eCommerce deja vu?

Identifying Winners, Losers, and the Unexpected Consequences

The recent attention around ChatGPT are well-deserved, I’ve played around with GPT4 and my god, it is remarkable. Yet, amidst all the excitement, there has been a noticeable lack of conversation surrounding the eerie resemblance of AI to the last eCommerce boom cycle. Maybe it’s because I’ve lived through the last one, that it feels like deja vu all over again…

The Marginal Cost of Software → Zero

When a variable approaches zero in any equation, intriguing properties start to emerge, incentive dynamics start to shift around. This principle holds true across physics as it does in the realm of economics and market incentives.

Throughout history, several technology-driven inflection points have reshaped our markets. Easy example would be data storage. The cost of data storage dropped from around $400K per GB in the 1980s → $10K in the 1990s → $0.37 in the 2000s → $0.02 in the 2010s, and finally → <$0.01 per GB in 2020.

In relation to GPT4, we’re seeing the last mile of marginal cost to develop software cross a very important threshold. The cost to develop an app was $100K+ in the 1990s to what would cost a few thousand in the 2000s, and just $100-1000 with a Shopify or SquareSpace account with some a bit of customization. But, with GPT4, it feels like this cost is being driven closer and closer towards zero – requiring no more than a simple click of a button.

With ChatGPT, we're crossing a threshold where the inertia and cost to develop a piece of software is falling far below the opportunity of a small or niche idea.

Ideas where the opportunity might’ve been too “small” may no longer be the case.

Bloodbath of golfballs and dog leashes

Over the past five years, I've observed the commoditization of eCommerce. The boom in businesses stemmed from the fact that anyone can create an eCommerce store. Want to make dog leashes? No problem – Alibaba has you covered. Need to ship custom golf balls? FlexPort will handle it. Looking for a storefront to sell your weighted blankets? Shopify makes that easy. The cost of building an eCommerce business in the 2000s, which required a DevOps team and a design team to launch a site, dropped from $100K+ down to just $29.99/mo to get started.

The last 10 years was productizing and reducing the cost to launch an online store. Now, more generalized software development is also following a similar trajectory. In a world where your core product is commoditized, the winners are determined by whoever has the best distribution. Whoever could acquire the most number of customers were the winners.

This actually made eCommerce into a game of “who had the best marketing, brand and sales & partnerships”? Distribution determined winners and losers. This, in turn, made the platforms on which these players duke it out become the true winners. While there were occasional winners and losers in each product category, such as Casper in bedding and UNTUCKit in apparel. Even the winners were not spared from brutal battles with copycats chipping away at market share and dealing with the consequences of platform changes by the likes of Facebook and other ad platforms.

The true winners in eCommerce boom era were the platforms themselves, ranging from Alibaba and FlexPort to Facebook and Shopify. Similarly, I expect a handful of winners at the platform level for ChatGPT. However, for these platforms to emerge victorious, many AI apps will need to become users and losers.

Micro-Niche Services enters the Arena

As the marginal cost of software creation approaches zero, it opens up the ability to create hyper customized instances of any software for any vertical. This will likely lead to a proliferation of benefiting niche and vertical specific businesses. Namely: services and SMBs. Businesses that traditionally catered to a smaller total addressable markets (TAM) will now be able to afford software that was traditionally "too expensive" or "not feasible" due to the simple math around ROI.

A dentist or electrician who interacted by pure word-of-mouth and a Yelp page is now able to afford the cost of developing a website with a custom CRM and analytics + tracking around customers and payroll. Such software would've traditionally costed tens of thousands of dollars will likely cost as little as $15/month with GPT4. Traditional businesses that did not have a true digital footprint due to cost constraints to now could fully upgrade at a minimal cost. Why not be SEO optimized as a dentist? Why not issue a loyalty card to each customer as a bookkeeper? Why not send hyper customized renewal notices via your CRM as a gardener?

These businesses now have the opportunity to bring the same degree of custom software sophistication that technology startups enjoyed in the 2010s. In a world where service businesses become a lot more competitive, they might actually turn into the driving force behind distribution. As in a world where software is commoditized, relationships is king.

You might start to see a 2 -person accounting shop have:

• Their custom website

• Automated custom onboarding with their own chatbot

• High level of CRM and therefore follow up cadences

• Their own fintech or loyalty point system

Overall, this should help boost margins or expand revenue for service businesses.

In the last decade, commerce companies were been beamed into the Atlantis of the cloud world, competing globally, selling locally. In this next decade, we may see service businesses enter the arena too. While existing locally, they're now one step elevated in a realm of competing more globally.

They've now received the same gear as the eCommerce companies of 2020s and the SaaS companies of 2010s.

Conclusion

How the landscape will shake out with GPT4 is yet to be determined, but one thing is for certain - this economic shift in cost will create a multitude of ripple effects. We still don’t know all the 3rd order impacts that will come with this, but there is a surprisingly high degree of similarities between this and eCommerce.

Ironically, the introduction of GPT4 might actually commoditize the very sector that created them - software development. And it might put more power into the hands of distribution - sales, marketing, and brand. After all, in a commoditized world, personalization and distribution becomes king.

In a world where distribution is king, power actually reverts back to personalized relationships. This could create a whole industry of service businesses that can compete globally, while existing locally.

As we enter this new era of micro-niche services and hyper-competition, it is crucial to remember that humans decide with their emotions and rationalize with their spreadsheets. The landscape is changing rapidly, and those who can adapt and leverage these new opportunities will be the ones who thrive.

---